Linear Feature Extraction for SAR Image based on Gabor Filter Edge Detector

Y. Kong, J. Zhou, and Y. Zhan (PRC)


SAR; ROEWA; Speckle noise; Edge detection map;Gabor filter; Extraction goals


Due to the presence of speckle, which can be modeled as a strong, multiplicative noise, edge detection in synthetic aperture radar (SAR) images is extremely difficult. In this paper, a new step-edge detector for SAR images is proposed, which is search of the best local adaptive Gabor filter based on ROEWA. Edge directions are determined by multi-direction characteristics of Gabor filter. At the same time, errors of edge directions are rectified by maximum likelihood estimation. By combination of the visual cells octave and correct edge direction, parameters of the best local Gabor filter are calculated for extracting the right edge in SAR images. Experiments show that the method achieves good effect edge detection, and extraction goals are more conform to the actual goal than ever, and has a strong versatility.

Important Links:

Go Back